Metagenomic profiling reveals lung multi-kingdom microbes as forensic markers for aquatic corpses investigation

The forensic investigation of corpses recovered from aquatic environments presents a major practical challenge. Recent studies have demonstrated that the bacterial community in the lung serves as a valuable indicator for diagnosing drowning, determining the drowning medium and estimating postmortem submersion interval (PMSI). However, the application and significance of lung multi-kingdom microbiome (archaea, eukaryota, and viruses) remains inadequately characterized. Meanwhile, the insufficient sequencing depth of commonly employed techniques, such as amplicon sequencing, restricts our understanding of microbial communities. In this study, we characterized the postmortem lung microbiome of mice submerged in water for up to 10 days using metagenomic sequencing, and subsequently validated the potential microbial biomarkers in both murine and human forensic specimens via qPCR. Integrated analyses were conducted followed by the confirmation of significant lung bacterial communities for drowning diagnosis, inference of drowning site, and estimation of the PMSI. Our findings revealed that bacteria constituted the predominant component of the lung microbiome in submerged murine carcasses, with eukaryota serving as the secondary dominant taxa. Seventeen bacterial and nine eukaryotic features at the species level were identified as potential biomarkers for drowning diagnosis. By detecting the specific molecular markers for Aeromonas species in both murine and human samples, the positive detection of Aeromonas species, particularly Aeromonas hydrophila, provides solid evidence for drowning diagnosis. Additionally, 14 and 17 bacterial species were identified as biomarkers for the inference of drowning site and estimation of PMSI, respectively. Based on the identified potential biomarkers, robust forensic models were constructed using the random forest (RF) algorithm. The accuracy of the bacterial model for drowning diagnosis was 89.29 %, while the accuracy of the eukaryotic model was 87.5 %. For the inference of the drowning site, the bacterial model achieved an accuracy of 100 %. Furthermore, the estimation of the PMSI yielded a mean absolute error of 0.66 ± 0.097 days. Collectively, our findings revealed that the selected 17 bacterial and 9 eukaryotic features in the lungs, particularly Aeromonas hydrophila, are beneficial for drowning diagnosis. Additionally, the other selected bacterial species contribute to the estimation of the drowning site and PMSI, thereby providing more comprehensive and refined information for accurate forensic investigations of corpses recovered from aquatic environments.